학술논문
An Experimental Evaluation in Plant Disease Identification Based on Activation-Reconstruction Generative Adversarial Network
Document Type
Conference
Author
Source
2022 2nd International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE) Advance Computing and Innovative Technologies in Engineering (ICACITE), 2022 2nd International Conference on. :361-366 Apr, 2022
Subject
Language
Abstract
Deep Neural network necessitates a big and specific plant disease data is scarce and also has equal structure, deep learning has recently shown promise in identifying plant lesions.” The data set must be supplemented with full-color plant lesion leaf images. This paper presents a Research on a strategy for obtaining a complete and unusual image of a plant disease leaf that could improve the classification network's accuracy. Among the benefits of our research are the following: We propose a binary generator network to address the question that how a generative opposite network (GAN) creates a disease image with a particular shape. and (ii) to tackle the problem of synthesizing a whole lesion leaf picture with multiple synthetic edge pixels and network out pixels using edge-smoothing and Image pyramid techniques. In the future, our method will show successfully increase plant disease Research and increase the categorization of network's accuracy. When compared to human experts and Alex Net, our method was shown to successfully enlarge the dataset of plant lesions and increase the accuracy of the classification network's identification